With all respect to Dfariswheel, (for whom I have a lot of respect), the Colt latch was not really designed for target shooters or to prevent accidentally opening the latch. It was a natural followup to the latch of the earlier Colt DA Navy (1889) and the Colt New Army and Navy. The latch on those guns contained a stop that kept the cylinder from rotating backward. To make that work, the latch had to pull back since it couldn't be pushed forward. That was, of course, a full decade before the S&W Model 1899, the first S&W to have a cylinder latch. (The Model 1896 opened by pulling the extractor rod forward.)
So Colt was in pioneering country, and that was their solution to a problem. Later, when they moved the sideplate to the left side, they had to incorporate the latch into the sideplate since they couldn't move it to the right side.
A good idea to have to pull the latch back? I don't think so, but I have used both types enough that I am not bothered by the Colt system. In fact, I prefer it to the Ruger or Dan Wesson system, though I agree with those who say the S&W is more instinctive and (IMHO) has no real problems.
As to Colt "stacking", that, too, is the result of the design; the highly skilled (and expensive) workers who tuned the Python actions only alleviated that condition; they did not eliminate it. The S&W cam design (which was incorporated into the M&P Model 1905 Third Change in 1909) operated to change the leverage of the trigger/hammer interface, eliminating the stacking that continued to plague Colt. Oddly, S&W seems not to have made much of that change; Neal and Jinks call it "a minor change ... to incorporate a slightly different double action throw". Pretty low key for a change that would eliminate stacking in the S&W and set it apart from almost every other DA revolver lockwork design!
Jim